Quality–Diversity and Illumination Algorithms in Discrete Combinatorial Domains: Diversity Metrics and Implications for Resilient Mining Operations

Quality–Diversity (QD) optimization has emerged as a distinctive paradigm in evolutionary computation, shifting the focus from identifying a single global optimum to illuminating a high-dimensional repertoire of elite solutions that jointly maximize performance and behavioral diversity. While algorithms like MAP-Elites have enabled transformative results in robotics and procedural content generation, their generalization to discrete combinatorial domains remains insufficiently consolidated in the literature. To address this gap, a systematic literature review was conducted strictly following PRISMA 2020 guidelines. The synthesis reveals rapid exponential growth in QD research, accompanied by significant algorithmic diversification toward gradient-informed variations and hardware-accelerated implementations. Despite this maturation, discrete combinatorial applications remain comparatively underrepresented, with only a small fraction (12.5%) of the analyzed corpus explicitly addressing discrete problems using domain-specific representations and heuristics. Based on these empirical findings, a conceptual framework is proposed. This framework positions QD as a vital mechanism for operational resilience in stochastic industrial contexts—specifically mining operations, including predictive maintenance, mineral processing optimization, and blast design—demonstrating its strategic value for complex decision-making.